Content uploaded by James Martin

Author content

All content in this area was uploaded by James Martin on Oct 04, 2019

Content may be subject to copyright.

ORIGINAL ARTICLE

J.C. Martin áW.W. Spirduso

Determinants of maximal cycling power: crank length,

pedaling rate and pedal speed

Accepted: 15 January 2001 / Published online: 15 March 2001

ÓSpringer-Verlag 2001

Abstract The purpose of this investigation was to de-

termine the eects of cycle crank length on maximum

cycling power, optimal pedaling rate, and optimal pedal

speed, and to determine the optimal crank length to leg

length ratio for maximal power production. Trained

cyclists (n=16) performed maximal inertial load cycle

ergometry using crank lengths of 120, 145, 170, 195, and

220 mm. Maximum power ranged from a low of 1149

(20) W for the 220-mm cranks to a high of 1194 (21) W

for the 145-mm cranks. Power produced with the 145-

and 170-mm cranks was signi®cantly (P<0.05) greater

than that produced with the 120- and 220-mm cranks.

The optimal pedaling rate decreased signi®cantly with

increasing crank length, from 136 rpm for the 120-mm

cranks to 110 rpm for the 220-mm cranks. Conversely,

optimal pedal speed increased signi®cantly with in-

creasing crank length, from 1.71 m/s for the 120-mm

cranks to 2.53 m/s for the 220-mm cranks. The crank

length to leg length and crank length to tibia length

ratios accounted for 20.5% and 21.1% of the variability

in maximum power, respectively. The optimal crank

length was 20% of leg length or 41% of tibia length.

These data suggest that pedal speed (which constrains

muscle shortening velocity) and pedaling rate (which

aects muscle excitation state) exert distinct eects that

in¯uence muscular power during cycling. Even though

maximum cycling power was signi®cantly aected by

crank length, use of the standard 170-mm length cranks

should not substantially compromise maximum power

in most adults.

Keywords Skeletal muscle áExercise test á

Cycling power áCrank length

Introduction

Previous investigators have reported that maximal

cycling power is aected by cycle crank length (Inbar

et al. 1983; Too and Landwer 2000; Yoshihuku and

Herzog 1990, 1996) and that the optimal crank length is

related to leg length (Inbar et al. 1983). Inbar et al.

(1983) reported that peak cycling power for the Wingate

anaerobic test varied by 8% for crank lengths of 125±

225 mm. Since Inbar et al. (1983) reported their ®ndings,

it has been reported that the resistance used in the

Wingate anaerobic test does not elicit maximum short-

term cycling power (Dotan and Bar-Or 1983; Patton

et al. 1985). Thus, their ®ndings regarding the eects of

crank length must be interpreted cautiously. Yoshihuku

and Herzog (1990, 1996) modeled mathematically the

eects of crank length on maximal power, optimal

pedaling rate, and optimal pedal speed. They reported

that maximum power varied by 0±10% for crank

lengths of 130±210 mm, depending on the de®nition of

optimal muscle length, and that optimal pedal speed

was nearly independent of crank length. Their model,

however, featured stepwise muscle activation and

relaxation and may not have been aected by reduced

muscle excitation, which normally occurs during acti-

vation and relaxation periods (Caiozzo and Baldwin

1997; Martin et al. 2000). More recently, Martin et al.

(2000) reported that impulse and power were similar for

crank lengths of 145±220 mm, but did not report values

for maximum power, optimal pedaling rate, or optimal

pedal speed. Thus, it seems that the exact eects of

crank length on maximum power, optimal pedaling rate,

and optimal pedal speed remain to be determined.

Therefore, the purposes of this investigation were to

determine the eects of crank length on the maximum

cycling power, optimal pedaling rate, and optimal pedal

Eur J Appl Physiol (2001) 84: 413±418

DOI 10.1007/s004210100400

J.C. Martin (&)

The University of Utah, Department of Exercise

and Sport Science, 250S. 1850E. Rm. 200, Salt Lake City,

UT 84112-0920, USA

E-mail: jim.martin@health.utah.edu

Tel.: +1-801-5877704

Fax: +1-801-5853992

J.C. Martin áW.W. Spirduso

The University of Texas at Austin, Department of Kinesiology

and Health Education, Austin, TX 78712, USA

speed of human subjects, and to determine the optimal

crank length for maximum power production.

Methods

Trained male cyclists [n=16, mean (SD) age: 29 (7) years, height:

179 (6) cm, mass: 73 (7) kg] volunteered to participate in this in-

vestigation. The procedures were explained and the subjects pro-

vided written informed consent to participate. This investigation

was approved by the Institutional Review Board at The University

of Texas at Austin.

Maximal cycling was performed using crank lengths of 120,

145, 170, 195, and 220 mm. Familiarization trials were performed

with all crank lengths during the week prior to data collection. On

each experimental data collection day, subjects performed a 5-min

warm-up of steady-state cycling at 100 W, and four maximal cy-

cling power tests at one crank length. A randomized counterbal-

anced design with four ordering sequences was used for the

presentation of the crank lengths to eliminate any presentation-

order eect.

Maximal cycling power was measured using the inertial load

method (Martin et al. 1997), which determines the torque and

power delivered to an ergometer ¯ywheel across a range of pedaling

rates. The ergometer was ®tted with bicycle-racing handlebars,

cranks, pedals, and seat, and was ®xed to the ¯oor. Each subject

wore cycling shoes ®tted with a cleat that locked into a spring-

loaded binding on the pedal.

A slotted disc was mounted on the ¯ywheel and an infra-red

photodiode and detector were mounted on the ergometer frame on

opposite sides of the disc. The slots were spaced at p/8 radians (Dh)

along the perimeter of the disc, and they alternately passed or in-

terrupted the infra-red light beam. The detector circuit was pro-

grammed to emit a square pulse at each interrupt. The time between

consecutive interrupts (Dt) was recorded by a dedicated micropro-

cessor with a clock accuracy of 0.5 ls. Flywheel angular velocity

was calculated as Dh/Dt. The time-angular velocity data were low-

pass ®ltered at 8 Hz using a ®fth-order spline (Woltring 1986).

Power for each revolution of the pedal cranks was calculated as the

rate of change in ¯ywheel kinetic energy for each complete revolu-

tion of the cranks (beginning with either leg). Maximum power was

identi®ed as the highest power for a complete revolution within each

bout (i.e., the apex of the power-pedaling rate curve). Pedaling rate,

in revolutions per minute (rpm), was calculated as the reciprocal of

the time (min) required to complete each revolution of the pedal

cranks. Pedal speed (PS; m/s) was calculated from pedaling rate

(PR; rpm) and crank length (CL; m) as: PS 2pPR CL=60.

Optimal pedaling rate and optimal pedal speed were de®ned as

those values at which maximum power occurred.

In the present investigation, speci®c changes were made to the

original protocol of Martin et al. (1997), including the length of the

crank, the inertial load (IL 0:5IGR2: where IL is the iner-

tial load, I is the moment of inertia of the ¯ywheel, and GR is the

gear ratio), and the number of crank revolutions. The inertial load

was varied by adjusting the gear ratio so that the inertial resistance

at the pedal (IRPIL=CL) was similar for all crank lengths

(Table 1). The inertial loads used in this investigation do not cor-

respond to outdoor cycling; rather, they were chosen speci®cally to

elicit maximum power during our power test. The number of crank

revolutions performed in each test was varied to match the total

work across all of the crank lengths (e.g., 6.5 revolutions for the

220-mm cranks; 9.0 revolutions for the 120-mm cranks). This ap-

proach allowed the subjects to reach optimal pedaling frequency

within approximately 2 s, and to complete the test in approximately

3±4 s. Seat height was set to match each subject's accustomed riding

position and was adjusted so that the distance from the top of the

saddle to the pedal axle (in its most extended position) was constant

for all crank lengths.

Leg length, femur length, and tibia length were recorded using a

®berglass measuring tape and an anthropometer. Leg length was

de®ned as the dierence between standing height and seated height.

Femur length was de®ned as the length from the greater trochanter

to the cleft of the knee joint. Tibia length was de®ned as the length

from the cleft of the knee to the lateral maleolus.

Repeated measures analysis of variance was used to determine

whether crank length signi®cantly aected maximum cycling

power, optimal pedaling rate, or optimal pedal speed. If signi®cant

(P<0.05) main eects of crank length were detected, the Bonferoni

post hoc procedure was used to determine which crank lengths

diered. Second-order polynomial regression analysis was per-

formed to determine the optimal crank length (as a ratio of leg

length, femur length, and tibia length) for maximum power. For

that regression analysis, maximum power for each test was scaled

as a proportion of that subject's maximum value for any crank

length. Data are presented as the mean (SEM), unless stated

otherwise.

Results

The maximum power of our subjects varied by 4%

across the range of crank lengths tested, from 1149

(44) W for the 220-mm cranks to 1194 (47) W for the

145-mm cranks (Fig. 1). Maximum power produced

when using the 145- and 170-mm cranks was signi®-

cantly greater (P<0.05) than that produced with the

120- and 220-mm cranks. Optimal pedaling rate de-

creased signi®cantly (P<0.05) with increasing crank

length (Fig. 2) from 136 (3) rpm for the 120-mm cranks

to 110 (3) rpm for the 220-mm cranks. The optimal

pedaling rate for the 195-mm cranks did not dier from

that of the 170- or 220-mm cranks, but the values for all

other lengths diered. Optimal pedal speed increased

signi®cantly (P<0.05) with increasing crank length,

Table 1 Ergometer settings for the ®ve crank lengths tested

Length Gear ratio Inertial load

(kgám

2

)

a

Pedal inertial

resistance (kgám)

120 mm 5.77:1 6.45 53.8

145 mm 6.38:1 7.88 54.4

170 mm 6.98:1 9.46 55.6

195 mm 7.59:1 11.2 57.3

220 mm 7.89:1 12.1 54.9

a

The inertial resistance at the pedal for the various crank lengths

do not match exactly because of the constraint of using bicycle

chain rings and cogs with integer numbers of teeth

Fig. 1 Maximum power. Power varied by 4%, and power

produced at the 145- and 170-mm cranks was greater than that

produced at the 120- and 220-mm cranks (*P<0.05)

414

from 1.71 m/s for the 120-mm cranks to 2.53 m/s for the

220-mm cranks; the value for each crank diered from

all others (Fig. 2).

Signi®cant second-order polynomial relationships

(P<0.001) were observed between power and crank

length relative to leg length [84 (4) cm], femur length [45

(2) cm], and tibia length [41 (3) cm]. The crank length to

leg length (Fig. 3) and crank length to tibia length ratios

accounted for 20.5% and 21.1% of the variability in

maximum power, respectively, whereas the crank length

to femur length ratio accounted for only 7.1% of the

variability. The optimal crank length for maximum

power was 20% of leg length or 41% of tibia length.

Discussion

The main ®ndings of this investigation were: (1) cycle

crank lengths that varied by 83% elicited a mere 4%

variation in maximum cycling power, (2) optimal ped-

aling rate decreased with increasing crank length,

whereas optimal pedal speed increased with increasing

crank length, and (3) the optimal crank length for

maximum power was 20% of leg length or 41% of tibia

length. The variation in maximum power was only half

as large as the 8% reported by Inbar et al. (1983), and

was within the range predicted by the mathematical

models of Yoshihuku and Herzog (1990, 1996). Part of

the dierence between the 4% variation in maximum

power in the present investigation and the 8% variation

in peak power reported by Inbar et al. (1983) may be due

to dierences in measurement techniques. As mentioned

earlier, it has been shown that the Wingate anaerobic

test does not elicit maximum power (Dotan and Bar-Or

1983; Patton et al. 1985). Rather, the standard Wingate

anaerobic test resistance (75 g/kg) allows subjects to

reach pedaling rates that are on the descending limb of

the power/pedaling rate relationship, where small dif-

ferences in pedaling rate may have a large eect on

power. In contrast, in the present investigation, the in-

ertial load method was used to determine the apex of the

power/pedaling rate relationship, and thus, our values

truly represent maximum cycling power for each crank

length. In addition, Inbar et al. (1983) used a single re-

sistance for the various crank lengths tested and thus

varied the resistance at the pedal. For example, a

Monark ergometer, with a resistive load of 5.25 kg (i.e.,

a load of 0.075 kg/kg body mass for a 70-kg subject) will

produce a resistive force of 289 N at the pedal for 170-

mm cranks, 410 N for 120-mm cranks, and 224 N for

220-mm cranks. Thus, by using a constant ¯ywheel re-

sistance for various crank lengths, Inbar et al. (1983)

altered dramatically the resistive force at the pedal. Our

solution to this interaction between crank length and

pedal force was to equate the ``inertial resistance at the

pedal'' (i.e., inertial load divided by the crank length) for

the various crank lengths tested. Thus, our methods

varied the resistive torque in order to hold constant the

resistive force at the pedal.

The model of Yoshihuku and Herzog (1990, 1996)

predicted that crank lengths ranging from 130 to 210 mm

would elicit 0±10% variation in maximum power, de-

pending on the de®nition of optimal muscle length.

When optimal muscle length was de®ned as the average

of each whole muscle's length during the cycle

(Yoshihuku and Herzog 1990, 1996: model 1), predicted

power varied by less than 1%. When optimal ®ber

length was based on the cross-bridge theory (Yoshihuku

and Herzog 1996: models 2a and 2b), power varied by

approximately 10%. The 4% variation observed in the

present data falls within the predictions of those two

models, and the power produced by our subjects (1149±

1194 W) was similar to the power predicted by the model

(922±1284 W for two legs).

Our results demonstrate that optimal pedaling rate

decreases with increasing crank length, whereas optimal

pedal speed increases with increasing crank length

(Fig. 2). Even though both of these variables are rate

Fig. 3 Maximum power versus leg length to crank length ratio.

The relationship of maximum power (expressed as a percent of

each subjects best performance) with the crank length to leg length

ratio (CL/LL) was parabolic, and the regression equation was:

proportion of maximum power = )6.83(CL/LL)2+ 2.77CL/LL

0:698; P<0.001, R

2

=0.205, SE=2.4%. The optimal crank length

was 20% of leg length

Fig. 2 Optimal pedaling rate and optimal pedal speed. Optimal

pedaling rate (j) decreased with increasing crank length (*diers

from all other crank lengths; **diers from all lengths except

195 mm; ***diers from the 120- and 145-mm crank lengths).

Optimal pedal speed (p) increased with increasing crank length and

the values for all cranks diered (*)

415

terms, they may represent distinct physiological phe-

nomena. Speci®cally, pedal speed constrains the short-

ening velocity of uniarticular muscles (Martin et al.

2000; Yoshihuku and Herzog 1990), whereas pedaling

rate aects muscle excitation (Caiozzo and Baldwin

1997; Martin et al. 2000). Caiozzo and Baldwin (1997)

reported that the incomplete excitation associated with

activation and relaxation kinetics reduced force output,

and that those kinetics exerted increasingly greater eect

at higher frequency. Consequently, average excitation

for a complete cycle was reduced with increasing fre-

quency. Thus, our data suggest that the optimal condi-

tions for maximum cycling power interactively depend

upon crank length, muscle shortening velocity (con-

strained by pedal speed), and muscle excitation state

(in¯uenced by cycle frequency).

The interactive eects of crank length on optimal

pedaling rate and pedal speed also extend to the entire

power/velocity relationship. Speci®cally, the power/

pedaling rate relationships for all ®ve crank lengths

(Fig. 4A) were similar in shape, but the relationships for

the shorter cranks were shifted to the right (higher

pedaling rate). The power/pedal speed relationships

were also similar in shape (Fig. 4B), but the relationships

for the longer cranks were shifted to the right (higher

pedal speed). The combined eects of pedaling rate and

pedal speed can be accounted for by using the product of

the two as an expression of ``cyclic velocity'' (Martin

et al. 2000). When power was plotted against cyclic

velocity, the relationships for all ®ve cranks tended to

become aligned (Fig. 4C), suggesting that pedal speed

(muscle shortening velocity) and pedaling rate (muscle

excitation state) interactively constrain muscular power

across a wide range of pedaling rates and pedal speeds.

Our results for optimal pedaling rate and optimal

pedal speed contrast with those predicted by Yoshihuku

and Herzog (1990, 1996). Their model predicted that

optimal pedal speed would be nearly constant for crank

lengths of 130±210 mm (2.5±2.8 m/s), but that optimal

pedaling rate would vary by over 100% (110±232 rpm).

As mentioned previously, their model featured stepwise

activation and deactivation and therefore was not sen-

sitive to the reduced muscle excitation associated with

increasing pedaling rate. Yoshihuku and Herzog (1990)

acknowledged that particular limitation of their model,

but suggested that it would mainly aect power at very

high pedaling rates. Our results suggest that the eect of

pedaling rate on muscle excitation is more pervasive and

aects power across a wide range of pedaling rates.

The selection of optimal cycle crank length for max-

imal power production may be of interest to competitive

cyclists and to researchers who use cycle ergometry as a

laboratory-based performance measure. Our data dem-

onstrate that the optimal crank length for maximal

power was 20% of leg length or 41% of tibia length. For

our subjects, the mean optimal crank length calculated

as a proportion of leg length [169 (2) mm] was similar to

that calculated as a proportion of tibia length [170

(3) mm]. Both of these are quite similar to the standard

length of bicycle and ergometer cranks (170 mm). Op-

timal crank length (i.e., 20% of leg length) varied from

151 mm for our shortest-legged (75.7 cm leg length)

subject to 183 mm for our longest-legged subject (91.4 cm

Fig. 4A±C Power/pedaling rate and power/pedaling speed rela-

tionships. The power/pedaling rate relationships (A) for all crank

lengths (j120 mm, p145 mm, u170 mm, )195 mm, m220 mm)

were similar in shape, but the relationships for the shorter cranks

were shifted to the right (i.e., toward a higher pedaling rate). The

power/pedal speed relationships were also similar in shape (B), but

the relationships for the longer cranks were shifted to the right (i.e.,

toward a higher pedal speed). When power was plotted against the

product of pedaling rate and pedal speed (``cyclic velocity'',

expressed as Hz ´m/s; C), the relationships for all ®ve cranks

tended to converge onto one curve

416

leg length). Even though the range in optimal crank

length of our subjects was 32 mm, the regression equa-

tion (Fig. 3) indicated that standard (170 mm) length

cranks would reduce power by less than 0.5%. Thus,

standard laboratory or bicycle equipment should not

substantially compromise maximum power for most

adults.

The optimal length determined from this investiga-

tion agrees well with that reported by Inbar et al. (1983:

166 mm). Indeed, even though the methods employed by

Inbar et al. (1983) were quite dierent from those used in

the present study, the results are qualitatively similar. A

seemingly major dierence, however, is the reported

correlation of maximum power with the leg length to

crank length ratio. Inbar et al. (1983) reported a corre-

lation coecient of 0.99, whereas our value was 0.45.

That dierence is due, at least in part, to the fact that

Inbar et al. performed regression on the mean values for

each crank length, whereas values for all 16 subjects

were included in our regression model. Indeed, when

similarly analyzed, the present data yield a correlation

coecient of 0.94. This dierence in analytical tech-

niques has important implications. The analysis report-

ed by Inbar et al. (1983) suggests that the crank length to

leg length ratio accounted for approximately 98% of the

variation in peak power, and that selection of optimal

crank length is essential for maximum power produc-

tion. Conversely, our analysis suggests that the same

ratio accounted for only 20.5% of the variation in

maximum power. Thus, even though our maximum

power values were highly reproducible [coecient of

variation = 1.8 (0.2)%], the crank length to leg length

ratio accounted for only one-®fth of the total variation.

This suggests that the selection of crank length will have

only a minor impact on maximal power production.

In most of the models presented by Yoshihuku and

Herzog (1990, 1996), the highest power was predicted for

the 130-mm crank, suggesting that their model may have

been more sensitive to muscle force/length characteristics

than our human subjects. That is, predicted power was

reduced by crank lengths that elicited muscle excursion

beyond the optimal portion of the force/length rela-

tionship. Their model included force/length eects based

on the model of Woittiez et al. (Woittiez et al. 1984). For

cycling, however, the muscles undergo cyclic length

changes and are stretched prior to each contraction.

Data reported by Neptune et al. (Neptune and Herzog

2000) suggest that some portion of that stretch may occur

after muscle activation. Speci®cally, they reported that

muscle electromyographic burst onset for the gluteus

maximus occurred as early as 54°before the top dead

center of the cranks (i.e., during the leg ¯exion phase)

when subjects pedaled at 120 rpm. The observation that

muscles that act to extend the leg are activated during

leg ¯exion suggest that they are subjected to active

stretch. Muscle force/length characteristics for cycling

may, therefore, be aected by stretch-enhanced force

production, which dramatically aects the force/length

characteristics. Speci®cally, Edman et al. (1978) reported

that muscle force was nearly constant from resting length

to 25% above resting length following stretch, whereas

force was reduced by approximately 15% over that same

range without stretch. In addition, Stevens (1993)

reported that muscles produced more force when per-

forming work-loops (activated 54°before shortening)

than when performing traditional force/velocity mea-

surements at the same velocity. Taken together, these

investigations provide compelling support for the notion

that stretch-enhanced force production aects muscular

power during cycling. Thus, stretch-enhanced force

production may account for at least some of the dier-

ences between our human data and the values predicted

by the model of Yoshihuku and Herzog (1990, 1996).

A potential limitation inherent in our methods was

that, by changing the inertial load in concert with crank

length, we may have in¯uenced the measurement of

maximum power, optimal pedaling rate, or optimal

pedal speed. However, the inertial load method used in

this investigation relies solely on the reaction torque of

the ¯ywheel to provide resistance. Thus, at any pedaling

rate, the measured power is exactly what the subject

produced, regardless of the inertial load. In addition,

extensive pilot testing in our laboratory has revealed

that both the maximum power and optimal pedaling

rate for 170-mm cranks were stable across a wide range

of inertial load conditions. Consequently, we are con®-

dent that the methods used in this investigation allowed

us to determine accurately maximum power, optimal

pedaling rate, and optimal pedal speed for each crank

length. Finally, the inertial loads used in this investiga-

tion were substantially lower than those experienced by

a cyclist using a racing gear ratio and, thus, might not

apply to road or track cycling. However, Fregly et al.

(1996) reported that inertial load has little eect on

pedaling coordination, suggesting that our ergometer

results are indeed applicable to outdoor cycling, even

though the inertial characteristics dier.

In summary, cycle crank lengths that varied by 83%

elicited a mere 4% variation in maximum cycling power.

Optimal pedaling rate decreased with increasing crank

length, whereas optimal pedal speed increased with in-

creasing crank length. The diering optimal conditions

for these two rate terms suggest that pedal speed (which

represents muscle shortening velocity) and pedaling rate

(which in¯uences muscle excitation) exert distinct eects

that limit muscular power during cycling. The optimal

crank length for maximal power was 20% of leg length

[169 (2) mm] or 41% of tibia length [170 (3) mm]. Even

though our results reveal an optimal crank length, it

must be recognized that the crank length to leg length

and crank length to tibia length ratios accounted for

only 20.5% and 21.1% of the variability in maximum

power exhibited by our subjects. Indeed, the use of

170-mm cranks would only reduce the power of our

shortest- and longest-legged subjects by less than 0.5%,

suggesting that standard laboratory or bicycle equip-

ment should not substantially compromise maximum

power for most adults.

417

Acknowledgements The authors wish to extend their sincere ap-

preciation to the participants in this study for their enthusiasm.

Experiments conducted in this investigation comply with current

laws in the United States of America.

References

Caiozzo VJ, Baldwin KM (1997) Determinants of work produced

by skeletal muscle: potential limitations of activation and re-

laxation. Am J Physiol 273:C1049±1056

Dotan R, Bar-Or O (1983) Load optimization for the Wingate

anaerobic test. Eur J Appl Physiol 51:409±417

Edman KA, Elzinga G, Noble MI (1978) Enhancement of me-

chanical performance by stretch during tetanic contractions of

vertebrate skeletal muscle ®bres. J Physiol (Lond) 281:139±155

Fregly BJ, Zajac FE, Dairaghi CA (1996) Crank inertial load has

little eect on steady-state pedaling coordination. J Biomech

29:1559±1567

Inbar O, Dotan R, Trousil T, Dvir Z (1983) The eect of bicycle

crank-length variation upon power performance. Ergonomics

26:1139±1146

Martin JC, Wagner BM, Coyle EF (1997) Inertial-load method

determines maximal cycling power in a single exercise bout.

Med Sci Sports Exerc 29:1505±1512

Martin JC, Brown NA, Anderson FC, Spirduso WW (2000) A

governing relationship for repetitive muscular contraction.

J Biomech 33:969±974

Neptune RR, Herzog W (2000) Adaptation of muscle coordination

to altered task mechanics during steady-state cycling. J Biomech

33:165±172

Patton JF, Murphy MM, Frederick FA (1985) Maximal power

outputs during the Wingate anaerobic test. Int J Sports Med

6:82±85

Stevens ED (1993) Relation between work and power calculated

from force-velocity curves to that done during oscillatory work.

J Muscle Res Cell Motil 14:518±526

Too D, Landwer GE (2000) The eect of pedal crank arm length

on joint angle and power production in upright cycle ergome-

try. J Sports Sci 18:153±161

Woittiez RD, Huijing PA, Boom HB, Rozendal RH (1984) A

three-dimensional muscle model: a quanti®ed relation between

form and function of skeletal muscles. J Morphol 182:95±113

Woltring HJ (1986) A FORTRAN package for generalized, cross-

validatory spline smoothing and dierentiation. Adv Eng Soft

8:104±113

Yoshihuku Y, Herzog W (1990) Optimal design parameters of the

bicycle-rider system for maximal muscle power output.

J Biomech 23:1069±1079

Yoshihuku Y, Herzog W (1996) Maximal muscle power output in

cycling: a modelling approach. J Sports Sci 14:139±157

418